Changes/additions to input character set handling:

 - After discussion with Simon, change the default input charset back to ASCII,
   rather than trying to work it out from the locale, for the sake of promoting
   .but file portability.

 - Add a new command-line option "--input-charset=csname", which overrides the
   ASCII default for all input files (since there's no other way to use a
   non-ASCII-compatible input file).

 - Output a warning if -Cinput-charset:foo is specified that it has no effect.

 - Update the docs to match all this. Also try to clarify some other things in
   this area that caught me out.

[originally from svn r5332]

1 files changed, 11 insertions, 0 deletions

diff --git a/error.c b/error.c
index 7b9ceaa..0011b18 100644
--- a/error.c
+++ b/error.c
@@ -37,6 +37,17 @@ static void do_error(int code, va_list ap) {
 	sprintf(error, "unrecognised option `-%.200s'", sp);
 	flags = PREFIX;
 	break;
+      case err_cmdcharset:
+	sp = va_arg(ap, char *);
+	sprintf(error, "character set `%.200s' not recognised", sp);
+	flags = PREFIX;
+	break;
+      case err_futileopt:
+	sp = va_arg(ap, char *);
+	sp2 = va_arg(ap, char *);
+	sprintf(error, "warning: option `-%s' has no effect%s", sp, sp2);
+	flags = PREFIX;
+	break;
       case err_noinput:		       /* no arguments */
 	sprintf(error, "no input files");
 	flags = PREFIX;